The defining fact of clinical interpreting is that every call is dense — there's no easy one to recover on. Here's how to run a live transcription and translation layer beside phone and video assignments, privately.
A patient intake call, then a bedside-adjacent video assignment for discharge instructions, then a pharmacy consult — interpreting for AMN Healthcare Language Services means every call is clinical, with no lighter call in between to recover on. A live captioning layer doesn't lighten the terminology — it gives you a second place to check a dosage or a name before you say it. This page is a practical guide to running one beside phone and video clinical work, written for the interpreter, not the agency.
AMN Healthcare Language Services routes interpreters into clinical settings almost exclusively — patient intake, bedside-adjacent rounds and discharge instructions over video, pharmacy consults, behavioral health sessions, and telehealth-style visits, along with a video roster that includes ASL and Certified Deaf Interpreter work alongside spoken languages. Phone assignments tend to connect on demand, queue-style; video assignments tend to run more like a scheduled clinical shift tied to a facility's own rounds.
In a broad OPI queue, the load comes from switching domains. In clinical interpreting, the domain never changes — but the density never lets up either. Drug names, dosages, procedure names, and diagnoses arrive at a pace and precision that leaves little slack, and a single mis-rendered number carries more weight here than in almost any other setting.
Video assignments add a second layer on top of that: staying visibly composed on camera while relaying something difficult to a patient who may be frightened, confused, or in pain. That combination — clinical precision plus visible emotional labor — is a specific kind of fatigue that a phone-only shift doesn't produce.
| Phone assignments (OPI) | Video assignments (VRI) | |
|---|---|---|
| Audio or video | Audio only — you never appear on screen | On camera — the patient sees you, and framing and composure are part of the job |
| Pace | Connects on demand, queue-style | Runs closer to a scheduled clinical shift tied to a facility's rounds |
| What's on your screen | Just the call and any notes you keep | The patient's video feed, plus anything else you have open on your side |
Unicaption is a real-time captioning and translation copilot that runs on your computer, beside whatever assignment is on today's schedule. It listens to your microphone or your system audio — it never dials in, never joins a video call, and never appears to the patient or the facility:
If you're weighing copilots built specifically for medical interpreting, our honest comparison with Interpreter maps where each one fits.
Confirm your audio source once — mic if calls come through a headset, system audio if the platform plays through your computer speakers. On a video assignment, position the caption window somewhere it won't pull your eyes off the patient's feed.
Interpret the way you always do — the transcript is a private backup, not a script. Lean on it for what's hardest to hold in memory under clinical pressure: dosages, drug names, and numbers. Our guide to Spanish-English medical interpreting covers the terminology side of this in more depth.
Reset before the next one — clinical interpreting rarely gives you a lighter call to recover on, so the gap between assignments is doing real work.
Add any new drug names or facility phrasing to your dictionary while they're fresh. A term captured once is a term you never have to hold in memory again.
Nearly everything that crosses your line in this work is protected health information — a diagnosis, a treatment plan, a mental health disclosure. HIPAA isn't background context here; it's the baseline every tool used beside the call has to meet.
Unicaption is end-to-end encrypted with nothing retained after the session: sessions are end-to-end encrypted, audio is never stored, transcripts are deleted the moment the session ends, and session content is never used to train models. On a video assignment specifically, it runs in a separate window on your device and never touches or appears in the video call itself.
That depends on AMN's own policies, which you should check directly — we won't pretend to speak for them. What Unicaption does is run locally beside your assignment, reading only your microphone or system audio; it never joins or records anything, never stores audio, and deletes transcripts when the session ends. It is end-to-end encrypted with nothing retained after the session. Whether an aid is appropriate for a given assignment is a professional judgment that stays with you.
No. Unicaption runs in a separate window on your own device and reads only your microphone or system audio — it never joins, records, or appears in the video call the patient sees. It's still worth confirming with AMN's own guidance on tools during video assignments, since expectations can differ from phone work.
Yes — it reads microphone or system audio identically for either. The only difference is that on a video assignment it stays entirely on your side of the screen; it never appears in the feed the patient or facility sees.
Every account gets 30 free minutes each week, no credit card required, and the free minutes renew weekly. Paid plans run from Instant at $9/month to Ultra at $149/month, with annual discounts. Spend a few weeks of free minutes on real phone and video assignments before paying anyone.
30 free minutes every week, no credit card. Nothing joins the call, nothing appears on screen but yours.
Start Free Trial →